Stereochemistry and conformation of skyllamycin, a non-ribosomally synthesized peptide from streptomyces sp. Acta 2897

Article abstract of DOI:10.1002/chem.201304562

Skyllamycin is a non-ribosomally synthesized cyclic depsipeptide from Streptomyces sp. Acta 2897 that inhibits PDGF-signaling. The peptide scaffold contains an N-terminal cinnamoyl moiety, a β-methylation of aspartic acid, three β-hydroxylated amino acids and one rarely occurring α-hydroxy glycine. With the exception of α-hydroxy glycine, the stereochemistry of the amino acids was assigned by comparison to synthetic reference amino acids applying chiral GC-MS and Marfey-HPLC analysis. The stereochemistry of α-hydroxy glycine, which is unstable under basic and acidic conditions, was determined by conformational analysis, employing a combination of data from NOESY-NMR spectroscopy, simulated annealing and free MD simulations. The simulation procedures were applied for both R- and S-configured α-hydroxy glycine of the skyllamycin structure and compared to the NOESY data. Both methods, simulated annealing and free MD simulations independently support S-configured α-hydroxy glycine thus enabling the assignment of all stereocenters in the structure of skyllamycin and devising the role of two-component flavin dependent monooxygenase (Sky39) as S-selective.

Full text of DOI:10.1002/chem.201304562

DOI: 10.1002/chem.201304562
Full Paper
&
Peptides
Stereochemistry and Conformation of Skyllamycin, a Non-
Ribosomally Synthesized Peptide from Streptomyces sp. Acta 2897
Vivien Schubert,[a] Florent Di Meo,[b, c] Pierre-Loıc Saaidi,[a, d] Stefan Bartoschek,[e] Hans-
¨
Peter Fiedler,[f] Patrick Trouillas,*[b, g, h] and Roderich D. Sꢀssmuth*[a]
Abstract: Skyllamycin is a non-ribosomally synthesized cyclic
depsipeptide from Streptomyces sp. Acta 2897 that inhibits
PDGF-signaling. The peptide scaffold contains an N-terminal
cinnamoyl moiety, a b-methylation of aspartic acid, three b-
hydroxylated amino acids and one rarely occurring a-hy-
droxy glycine. With the exception of a-hydroxy glycine, the
stereochemistry of the amino acids was assigned by compar-
ison to synthetic reference amino acids applying chiral GC-
MS and Marfey-HPLC analysis. The stereochemistry of a-hy-
droxy glycine, which is unstable under basic and acidic con-
ditions, was determined by conformational analysis, employ-
ing a combination of data from NOESY-NMR spectroscopy, si-
mulated annealing and free MD simulations. The simulation
procedures were applied for both R- and S-configured a-hy-
droxy glycine of the skyllamycin structure and compared to
the NOESY data. Both methods, simulated annealing and
free MD simulations independently support S-configured a-
hydroxy glycine thus enabling the assignment of all stereo-
centers in the structure of skyllamycin and devising the role
of two-component flavin dependent monooxygenase
(Sky39) as S-selective.
Introduction
PDGF-signaling plays a crucial role in the development of vari-
ous diseases including cancer, fibrosis or arteriosclerosis,[1,2]
and therefore PDGFs and their receptors constitute important
targets for the development of anti-cancer drugs.[3] While the
marketed drugs Imatinib and Sorafenib address the ATP-bind-
ing pocket of the PDGF receptor and other protein kinases, the
cyclic depsipeptide skyllamycin has been found to inhibit the
binding of platelet-derived growth factor BB (PDGF BB) to its
receptor PDGF b, thus representing an alternative in the inhibi-
tion of PDGF signaling.
[a] Dipl.-Chem. V. Schubert, Dr. P.-L. Saaidi, Prof. Dr. R. D. Sꢀssmuth
Institut fꢀr Chemie, TU Berlin
Straße des 17, Juni 124, 10623 Berlin (Germany)
E-mail: suessmuth@chem.tu-berlin.de
[b] Dr. F. Di Meo, Prof. Dr. P. Trouillas
INSERM UMR850, Facultꢁ de Pharmacie, Universitꢁ de Limoges
2 rue du Dr. Marcland, 87025 Limoges (France)
E-mail: patrick.trouillas@unilim.fr
[c] Dr. F. Di Meo
Department of Physics, Chemistry and Biology (IFM)
Linkçping University, SE-58183 Linkçping (Sweden)
Skyllamycin A was firstly isolated as antibiotic RP-1776 from
Streptomyces sp. KY 11784,[4] and subsequently found as a two
component mixture together with skyllamycin B, from Strepto-
myces sp. Acta 2897.[5] Skyllamycin A and B (Scheme 1) are
non-ribosomally synthesized cyclic 11mer depsipeptides with
a 2-[1-(Z)-propenyl]-cinnamoyl (PrnCin) moiety attached to the
N-terminus. A preliminary structural characterization of the
11mer skyllamycin A peptide scaffold reveals a b-methylation
of aspartic acid (b-Me-Asp3) as well as a remarkably high con-
tent of hydroxylated amino acids. Namely, three amino acids
are b-hydroxylated (b-OH-Phe5, b-OH-O-Me-Tyr7 and b-OH-
Leu11) and one glycine is hydroxylated (a-OH-Gly9). b-Hydroxy
amino acids are present in various antibacterial substances as
for example vancomycin,[6] ramoplanin,[7] or lysobactin,[8] while
a-OH-Gly was only found in the antitumor antibiotic spergua-
lin.[9]
[d] Dr. P.-L. Saaidi
Universitꢁ Evry Val d’Essonne
Boulevard FrannÅois Mitterand, 91000, Evry, France and
CNRS-UMR8030, 2 rue Gaston Crꢁmieux, 91057 Evry, France
and
CEA, DSV, IG, Genoscope, 2 rue Gaston Crꢁmieux, 91057 Evry, France
[e] Dr. S. Bartoschek
Sanofi, Research & Development, S&I - Strategy, Science Policy &
External Innovation, Industriepark Hoechst
Bldg. H831, 65926 Frankfurt am Main (Germany)
[f] Prof. Dr. H.-P. Fiedler
Mikrobiologisches Institut, Eberhard-Karls-Universitꢂt Tꢀbingen
Auf der Morgenstelle 28, 72076 Tꢀbingen (Germany)
[g] Prof. Dr. P. Trouillas
Service de Chimie des Matꢁriaux Nouveaux, Universitꢁ de Mons-UMONS
Place du Parc 20, 7000 Mons (Belgique)
[h] Prof. Dr. P. Trouillas
Regional Centre of Advanced Technologies and Materials
Department of Physical Chemistry, Faculty of Science
Palacky University, tr. 17 listopadu 12, 777146 Olomouc (Czech Republic)
In Streptomyces sp. Acta 2897, the peptide chain of skyllamy-
cin is assembled by three non-ribosomal peptide synthetases
(Sky29, 30, and 31) consisting of 11 modules.[5] The adenylation
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201304562.
Chem. Eur. J. 2014, 20, 4948 – 4955
4948
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Enantiomer analytics of skyllamycin hydrolysates
The stereochemistry of all amino acids except for a-OH-Gly9
was determined employing the following workflow: 1) hydroly-
sis of skyllamycin A and B, 2) derivatization of the amino acids
followed by 3) chiral GC-MS and/or HPLC-MS analysis, and
4) comparison to reference amino acids. Accordingly, the con-
figuration of the proteinogenic amino acids was determined as
S-Thr1, S-Ala2, S-Pro6, R-Leu10 (Table 1). Under the hydrolytic
Table 1. Assignment of the amino acid configurations of skyllamycin A
by means of chiral GC-MS and HPLC-MS.
Residue
Config.
Chiral GC-MS
Marfey HPLC-MS
Scheme 1. Structure of cyclodepsipeptides skyllamycin A and B.
Thr1
2S,3R
S
2S,3S
2S,3S
S
2S,3S
R
–
+
+
+
+
+
+
À
À
+
+
+
+
À
À
+
À
+
À
+
+
Ala2
b-Me-Asp3
b-OH-Phe5
Pro6
b-OH-O-Me-Tyr7
Trp8
a-OHGly9
Leu10
b-OH-Leu11
domains of each module are specific for the incorporation of
each of the 11 amino acids. Furthermore, epimerization do-
mains in module 8 and module 10 indicate that Trp8 and Leu10,
respectively are R-configured in the peptide. A promiscuous
P450 monooxygenase introduces the three b-hydroxylations
(Sky32) and a two-component flavin dependent monooxyge-
nase (Sky39) is responsible for the a-hydroxylation.
R
2R,3S
Both the remarkable structural features and the bioactivity
of skyllamycin led us to commence a study on a comprehensive
assignment of skyllamycin stereochemistry and conformation.
This was achieved by combining data from enantiomer analyt-
ics of amino acids, with NOESY-NMR spectroscopy, simulated
annealing and free molecular dynamic (MD) simulations. The
molecular modeling simulations were applied with both S- and
R-configured a-OH-Gly9, with and without constraints issued
from NOESY-NMR data. This joint experimental and in silico
procedure provides a clear evidence for the absolute stereo-
chemistry of skyllamycin A.[10]
conditions that were applied (6 N HCl,1108C, 24 h), the degra-
dation of Trp was prevented by addition of dithiothreitol
(DTT). Unlike other amino acids, enantioseparation of Trp could
not be achieved using a chiral cyclodextrin phase used for GC-
MS analytics. Therefore samples were derivatized with Marfey’s
reagent (1-fluoro-2,4-dinitrophenyl-5 l-alanine amide)[10] and
analyzed by HPLC-MS to assign R-Trp as a constituent of skylla-
mycin (Table 1). In order to clearly identify the stereoisomers of
the non-proteinogenic b-hydroxy amino acids and b-Me-Asp,
we synthesized a set of stereoisomers as reference compounds
according to previously published procedures.[13–16] Hence, con-
figurations of the non-proteinogenic amino acids were as-
signed as follows: (2S,3S)-b-OH-Phe, (2R,3S)-b-OH-Leu and
(2S,3S)-b-Me-Asp (Table 1). Under acidic hydrolysis conditions
the b-OH-O-Me-Tyr constituent of skyllamycin was degraded
and no remains were detected by GC-MS analysis.[17,18] Skylla-
mycin A was therefore submitted to ozonolysis which, followed
by an oxidative workup, transforms b-OH-O-Me-Tyr into the
acid-stabile b-OH-Asp.[18] Following these lines, synthetic stand-
ards of b-OH-O-Me-Tyr were ozonolyzed in the same way. The
subsequent GC-MS analysis revealed the presence of (2S,3R)-b-
OH-Asp in the hydrolysate, and thus (2S,3S)-b-OH-O-Me-Tyr in
skyllamycin A.
Results
The considerable tailoring of the skyllamycin peptide namely
methylation and hydroxylation of amino acids in skyllamycin
(b-OH-Phe5, b-OH-O-Me-Tyr7, b-OH-Leu11, a-OH-Gly9, and b-Me-
Asp3) constitutes a significant challenge for a full stereochemi-
cal assignment. The main obstacle however concerns a-OH-
Gly9 that is unstable under basic and acidic conditions, thus
eluding from conventional amino acid analysis techniques.[5]
Unlike previous examples of peptide antibiotics in which crys-
tallization led to a full stereochemical assignment including
a 3D model,[11,12] such attempts were not successful for skylla-
mycin. Therefore, the configuration of all amino acids, except
for a-OH-Gly9, had to be analyzed by amino acid analysis. With
regard to the configuration of a-OH-Gly9, NOESY-NMR spectra
were employed to derive constraints for molecular modeling
simulations (both simulating annealing and molecular dynam-
ics). To reinforce the robustness and sampling of the conforma-
tional analysis, unconstrained simulations were performed and
the thereof derived distances were compared to forty NOESY-
NMR-derived distances.
Proton–proton distances of skyllamycin from NOESY-NMR
data
While NMR spectroscopy is capable to solve the relative stereo-
chemistry, only in combination with molecular modeling and
based on the knowledge of at least one remote stereocenter
the absolute stereochemistry can be determined. At this stage
Chem. Eur. J. 2014, 20, 4948 – 4955
www.chemeurj.org
4949
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 2. Proton–proton distances dA–B [ꢂ] determined from NOESY-NMR
spectra. Distances in italics were constrained for molecular modeling
simulations.
Proton pair
A
B
dA–B [ꢂ][a]
PrnCin-H3
PrnCin-H5
PrnCin-H10
PrnCin-H10
Thr1-NH
b-Me-Asp3-HMe
Leu10-H1
3.08
3.37
2.80
3.70
2.41
2.85 (d1)
3.49 (d2)
3.03
2.38
2.69
3.05
3.66
2.88
2.65
2.85
6.67
2.49
2.73
3.44
2.92
3.94
2.45
3.10
3.42
2.21
2.63
2.06
3.73
3.53
2.86
2.82
3.23
2.65
3.31
2.80
3.73
3.44
3.76
3.09
2.79
PrnCm-H3
b-Me-Asp3-HMe
Thr1-Hg
Thr1-Hg
a-OH-Gly9-Ha
a-OH-Gly9-NH
Ala2-Ha
Thr1-Hg
Ala2-NH
Ala2-NH
Ala2-Hb
Ala2-Hb
Ala2-Ha
Ala2-Hb
b-Me-Asp3-NH
b-OH-Leu11-Ha
b-Me-Asp3-Hb
b-Me-Asp3-HbMe
b-OH-O-Me-Tyr7-Ha
b-OH-O-Me-Tyr7-Hb
Trp8-H6
Ala2-Hb
b-Me-Asp3-NH
b-Me-Asp3-Hb
b-OH-O-Me-Tyr7-NH
Trp8-H1
Trp8-H5
Trp8-H5
Trp8-Ha
Trp8-H5
Trp8-Hb
Trp8-H5
Trp8-Hb1
Figure 1. Representative conformations of the peptide backbone with a-OH-
Gly (ball model) highlighted for a) S-configured (in agreement with NOESY-
NMR data; (conformer 3) and b) R-configured skyllamycin (conformer 6, see
text below). The d1 and d2 distances were chosen as being characteristic of
the intramolecular folding.
Trp8-H5
a-OHGly9-NH
Trp8-Hb
Trp8-NH
Trp8-NH
Trp8-Hb1
Trp8-NH
b-OH-O-Me-Tyr8-NH
b-OH-O-Me-Tyr8-Ha
Trp8-Hb1
Trp8-NH
Trp8-Ha
Trp8-Ha
a-OH-Gly9-NH
a-OH-Gly9-Ha
a-OH-Gly9-NH
a-OH-Gly9-NH
a-OH-Gly9-Ha
b-OH-Leu11-NH
Leu10-Hb
Trp8-Ha
a-OH-Gly9) as indicated in Figure 1and Table 2. These two dis-
tances, which appear adequately characteristic of the intramo-
lecular folding of skyllamycin A (Figure 1), were followed
within the resulting conformers of all simulations. They provide
information on the backbone architecture, since protons as-
signed to Thr1 and a-OH-Gly9 show a strong NOE-contact
(Table 2 and Supporting Information), while being distant in
a sequential context. Hence, only in silico conformers with
values of d1 and d2 close to the experimental data are likely to
resemble the natural conformer in solution.
Trp8-Hb
Trp8-Hb1
a-OH-Gly9-NH
a-OH-Gly9-H
Leu10-NH
Leu10-NH
Leu10-NH
Leu10-Ha
Leu10-Hb
Leu10-Hb1
b-OH-Leu11-Hb
b-OH-Leu11-Hb
Leu10-Hb1
a-OH-Gly9-Ha
a-OH-Gly9-Ha
b-OH-Leu11-NH
b-OH-Leu11-NH
b-OH-Leu11-Hd
b-OH-Leu11-Hd1
[a] The distance between Trp8-H5 and Trp8-H6 (dTrp5–6 =2.49) was used as
reference.
Simulated annealing of skyllamycin with S- and
R-configured a-OH-Gly9
of the structure elucidation, a-OH-Gly9 remained the only
amino acid for which the stereocenter was uncertain. Therefore
NOESY-NMR spectra of skyllamycin A were recorded in order to
identify inter-residue distances characteristic of both conforma-
tional and stereochemical features. From the NOESY-NMR-de-
rived signals forty proton–proton distances (dA–B) were ob-
tained corresponding to forty intra- and inter-residue contacts
(Table 2 and Supporting Information). Out of these, eleven con-
tacts were assigned to interactions directly involving the a-OH-
Gly9 moiety (Table 2). The eleven corresponding dA–B distances
were used when molecular modeling simulations were con-
strained (see below).
The simulated annealing procedure allows a complete explora-
tion of the entire conformational space. It was applied to both
S- and R-configured a-OH-Gly in skyllamycin A, and provided
60 conformers of each stereoisomer. These are grouped into
30 conformers from constrained and 30 from unconstrained
simulations (see below). When applied, the force constants of
the constraints were low enough to enable sufficient flexibility
for d1 and d2 in a range from 3 to 13 ꢂ (Figure 2).[19] As a result,
the distances d1 (2.85 ꢂ) and d2 (3.49 ꢂ) determined from
NOESY-NMR spectra could only be achieved for the S-configu-
ration at a-OH-Gly9 (Figure 2 and Table 3). In contrast for the
structure with R-configuration for all 60 conformers d1 and d2
were found in the range >4.5 ꢂ (Figure 2). In addition, when
comparing the NOESY-NMR-derived experimental distances to
the corresponding distances determined by simulated anneal-
To elaborate the configuration of a-OH-Gly, two distances
between both a-OH-Gly9 and Thr1 were considered: d1 (H-g of
(s)-Thr1 to H-a of a-OH-Gly9) and d2 (H-g of (s)-Thr1 to N-H of
Chem. Eur. J. 2014, 20, 4948 – 4955
www.chemeurj.org
4950
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
configuration, which perfectly fit to the experimental data de-
rived from NOESY-NMR spectra (Figure 2 and Table 4).
Free MD simulations of skyllamycin with S- and
R-configured a-OH-Gly9
In addition to simulated annealing, 100 ns free MD simulations
were performed with NOESY-NMR-derived distances to protons
of a-OH-Gly9 (Table 2) constrained during the first 10 ns.[20] In-
terestingly, after the first 10 ns, that is, an unconstrained mode,
the S-configured structure keeps the expected d1 and d2 dis-
tances along the rest of the MD simulation (Figure 3a), while
Figure 2. Distribution of intramolecular distances between H-g of Thr1 and
1) H-a of a-OH-Gly9 (d1, ꢂ) or 2) NH of a-OH-Gly9 (d2, ꢂ). Distances d1 and d2
were obtained from simulated annealing simulations for skyllamycin with
either R-configured (triangle, full or empty when obtained from NMR-con-
strained and unconstrained simulated annealing) or S-configured (square,
full or empty when obtained from NMR-constrained and unconstrained si-
mulated annealing) a-OH-Gly9. Experimental data (black dot) with a standard
error of +/- 0.5 (black cycle) are depicted.
Table 3. Comparison of distances d1–d2 (ꢂ) of S-(a-OH-Gly9) skyllamycin
conformers displaying the best fit to NOESY-NMR data: Conformer 1 from
simulated annealing and conformer 3 from free MD simulations.
d1
d2
simulated annealing (conformer 1)
Free MD simulations (conformer 3)
NOESY-NMR data
3.39
3.87
2.85
4.66
3.74
3.49
Figure 3. Evolution of interatomic distances d1 (left) and d2 (right) (ꢂ) be-
tween the protons of Thr1-Hg and a-OHGly9-Ha and a-OHGly9-NH, respec-
tively, along the MD simulation (NMR-constrained during the first 10 ns) for
skyllamycin with a) S- and b) R-configured a-OHGly9. The dashed lines show
the d1 and d2 distances derived from NOESY-NMR data.
Table 4. Mean absolute deviation (MAD, ꢂ), root-mean-square deviation
(RMSDtot, ꢂ) based on all 40 NOESY-NMR-derived distances or on 11 dis-
tances in proximity of a-OH-Gly (RMSD, ꢂ) of three conformers each (R-
and S-(a-OH-Gly9) skyllamyin A) with the best fit to NOESY-NMR data.
in the R-configured structure both distances d1 and d2 increase
up to 10 ꢂ (Figure 3b). The free MD simulations yield conform-
ers 3 and 6 for S- and R-configured a-OH-Gly9 skyllamycin (see
Tables S3 and S4 in the Supporting Information), respectively.
Conformer 3 has the same geometrical features as conformers
1 and 2 determined by simulated annealing and exhibits a flat-
tened shaped backbone (Figure 1). Likewise conformer 6
shares the same conformational features as conformers 4 and
5 which are of circular shape (Figure 1). Independently from
the simulated annealing procedure, this result corroborates
that the NOESY-NMR data (Table 4) strongly correlate to an S-
configured skyllamycin.
Stereoisomer
Geometry[a]
MAD [ꢂ]
RMSDtot [ꢂ]
RMSD [ꢂ]
S-configured
1[b]
2[b]
3[c]
4[b]
5[b]
6[c]
0.86
0.96
0.51
1.12
1.18
1.27
1.85
2.30
1.40
2.32
2.43
2.59
0.66
1.04
0.76
2.20
1.98
2.88
R-configured
[a] The xyz-coordinates of all six conformers are given in Tables S3 and S4
in the Supporting Information. [b] Conformers 1–2 and 4–5 are obtained
from simulated annealing calculations and are considered as having the
best agreement with the NOESY-NMR data. [c] Conformers 3 and 6 are
obtained after equilibration of the 100 ns-MD simulations (with constrains
inspired by NOESY-NMR-distances during the first 10 ns).
Discussion
The cyclic depsipeptide skyllamycin is a highly hydroxylated
peptide with enigmatic a- and b-hydroxylations, synthesized
by non-ribosomal peptide synthetases in interaction with tai-
loring enzymes.[5,21] With one exception, the herein presented
stereochemical assignment derived from the amino acid analy-
ing, the S-configured conformers exhibit much lower RMSD
(root mean square deviation) compared to those of the R-con-
figured skyllamycin (Table 4). From the simulated annealing
procedure we could identify two conformers (1 and 2) with S-
Chem. Eur. J. 2014, 20, 4948 – 4955
www.chemeurj.org
4951
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
ses is in agreement with those predicted by A domain specific-
ities and the domain arrangement in the non-ribosomal pep-
tide synthetases Sky29-31 of the skyllamycin biosynthesis gene
cluster.[5] From the amino acids with one chiral center only
Leu10 and Trp8 are R-configured, while the other amino acids
are S-configured in the a-position. Modifications in the b-posi-
tion of non-proteinogenic amino acids, like b-hydroxylation
and b-methylation show S-configuration. Only b-OH-Leu11 with
a (2R,3S)-configuration does not fit to the biosynthetic logic,
since no epimerization (E) domain in module 11 of the non-ri-
bosomal peptide synthetase Sky31 has been found.[5] However
such a case, that is, a non-ribosomal peptide synthetase lack-
ing modules with an epimerization domain while displaying R-
configuration in the product peptide, is not without prece-
dence as this is also observed for example, for vancomycin[22]
or cyclosporine.[23]
droxy groups of b-hydroxy amino acids (b-OH-Phe5, b-OH-Tyr7,
b-OH-Leu11) as well as of a-OH-Gly9. Specifically, these interac-
tions include the H-bonds between the b-hydroxyls of b-OH-
Phe5 with the carbonyl (C=O) of b-OH-O-Me-Tyr7; of b-OH-O-
Me-Tyr7 with C=O of Pro6; and of b-OH-Leu11 with C=O of
Leu10. This is complemented by the interaction of the a-OH (a-
OH-Gly9) with C=O of Trp8. The fifth hydrogen bond is contrib-
uted by the interaction between NH of Gly4 and C=O of Ala3.
Furthermore, the H-bond network displayed in Figure 4a pre-
dominantly favors interactions between the preceding and the
subsequent residues of the amino acid sequence. This finding
is in stark contrast to the structure of the R-isomer, which
shows hydrogen bonds between NH of Trp8 and C=O of b-OH-
Phe5, between b-OH of b-OH-Phe5 and C=O of Trp8 and be-
tween NH of Gly4 and C=O of b-OH-Leu11 and does not display
any H-bonds involving a-OH-Gly9. The loss of the hydrogen
bond between the a-OH (a-OH-Gly9) and C=O of Trp8, which is
a main difference between the S- to the R-configured confor-
mer, appears of particular importance as it leads to a dramatic
intramolecular reorganization, which impedes stabilization of
the hydrogen-bonding network.
The configurational assignment of the above-mentioned
amino acids was the basis for subsequent simulated annealing
and free MD simulations. Both methods independently favor
the S-configuration for a-OH-Gly for reasons of steric hindrance
and hydrogen-bonding capabilities. Accordingly a view at the
architecture of skyllamycin (Figure 4) shows an inter-residue
The herein presented findings also have implications for the
biosynthesis of skyllamycin: The glutamate mutase (Sky41, Sky
42) that catalyzes the conversion of glutamate to b-methylas-
partate is S-selective in the b-position. The b-hydroxylations
performed by the P450 monooxygenase (Sky32) are all S-selec-
tive and are introduced while the amino acids are bound to
the peptidyl carrier protein before further being coupled to
the peptide chain.[21] This could allow for formation of hydro-
gen bonds between the b-hydroxy groups and neighboring
carbonyls already before final macrolactonization by the thio-
esterase domain of Sky31. The hydrogen bonds could contrib-
ute to a preorganization of the peptide chain in order to facili-
tate the ring closure reaction. Finally, according to our data
also the flavin-dependent oxygenase Sky39, introducing the a-
hydroxy group into a-OH-Gly, is S-selective. The a-hydroxyl-
ation is used in primary metabolism for the processing of
many peptide hormones, neurotransmitters and growth factors
to C-terminal amides. There peptide amidation is catalyzed by
the bifunctional enzyme peptidylglycine a-amidating monoox-
ygenase (PAM).[24] The conversion is a two-step process involv-
ing the peptidylglycine a-hydroxylating monooxygenase
(PHM) and peptidyl a-hydroxyglycine a-amidating lyase (PAL).
PHM catalyzes stereoselectively the hydroxylation to S-a-OH-
Gly, whereas PAL stereoselectively induces the cleavage of the
NÀC bond in C-terminal S-a-OH-Gly.[24,25] PAL does not react
with R-a-OH-Gly.[25] Likewise, for the secondary metabolite
spergualin an S-configuration has been determined by use of
a serine carboxypeptidase.[26] Interestingly, the S-enantiomer of
15-deoxyspergualin showed anti-leucemic activity whereas the
R-enantiomer is almost inactive.[26] In skyllamycin the stability
of this modification and protection from hydrolysis may be
supported by its involvement into the extensive H-bond net-
work. With regard to the biosynthesis the time point of a-hy-
droxylation of Gly by Sky39 is still elusive. Hydroxylation of free
or PCP-bound Gly seems highly unlikely due to the required
stabilization of an aminal-like structure for subsequent peptide
Figure 4. Conformers from MD simulations (3 and 6) highlighting the hydro-
gen bond network for skyllamycin with a) S-configured and b) R-configured
a-OH-Gly9.
hydrogen-bonding network, which is significantly different in
both S- and R-configured skyllamycin (Figure 4), favoring a flat-
tened-type conformer and a circular-type conformer, respec-
tively (represented by conformers 3 and 6 respectively in
Figure 1 and Figure 4). For the S-isomer five strong hydrogen
bonds with bond distances in the range of 1.78-2.01 ꢂ (see Fig-
ure 4a) are observed, contrasted by only three in the range of
1.97-2.52 ꢂ (see Figure 4b) for the R-isomer. Interestingly, the
hydrogen bond network of the S-isomer accommodates all hy-
Chem. Eur. J. 2014, 20, 4948 – 4955
www.chemeurj.org
4952
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
GC-MS analytics 1 mL of this solution was applied to the injector.
The following temperature program was used: isothermal at 708C
for 2 min, gradient of 38CminÀ1 up to 908C, isothermal at 908C for
15 min, gradient of 108CminÀ1 up to 2008C, isothermal at 2008C
for 10 min. Retention times of the derivatized amino acids (min): R-
Ala (13.7), S-Ala (15.2), (2S,3R)-b-OH-Leu (18.0), (2R,3S)-b-OH-Leu
(20.6), R-Leu (21.2), S-Leu (25.5), S-Thr (26.1), (2R,3R)-b-OH-Leu
(26.7), R-Thr (27.1), R-aThr (27.5), (2S,3S)-b-OH-Leu (27.8), S-aThr
(28.5), S-Pro (30.0), R-Pro (30.7) R-Asp (31.0), S-Asp (31.3), (2R,3R)-b-
OH-Asp (32.9), (2S,3S)-b-OH-Asp (33.7), (2R,3S)-b-OH-Asp (34.9),
(2S,3R)-b-OH-Asp (35.3), (2S,3R)-b-MeAsp (38.2), (2S,3S)-b-MeAsp
(39.3), (2S,3R)-b-OH-Phe (43.2), (2R,3S)- b-OH-Phe (43.2), (2R,3R)-b-
OH-Phe (45.9), (2S,3S)-b-OH-Phe (46.7).
coupling. Therefore we suggested this reaction to occur either
on the NRPS-bound peptide or after NRPS assembly and re-
lease from the NPRS assembly line.
Conclusion
In summary, a combination of bioanalytical methods and calcu-
lations succeeded in a full stereochemical assignment of the
PDGF inhibitor peptide skyllamycin with a particular focus on
the configuration of the unusual amino acid a-OH-Gly. These
tailoring modifications performed by the biosynthetic assembly
line of skyllamycin, consisting of b-methylation of Asp, b-hy-
droxylations of Phe, Tyr and Leu as well as a-hydroxylation of
Gly, are S-selective. The full conformational analysis describing
intramolecular rearrangements confirms the NOESY-NMR con-
tacts experimentally observed. While this work illustrates gain-
ing data on conformation and the stereochemistry even of
flexible molecules by combining MD simulations with NOESY-
NMR data this structure elucidation will aid in using skyllamy-
cin as a lead structure for the design of novel PDGF inhibitor
molecules.
Ozonolysis of b-hydroxy-O-methyl tyrosine and of
skyllamycin
Amino acid (1 mg) or skyllamycin (1 mg) was dissolved in MeOH
(1 mL) and cooled down to À788C. A stream of ozone was bub-
bled for 75 min through the solution. Subsequently 440 mL of H2O2
(30%) were added to the reaction mixture which was then allowed
to stand at room temperature for 12 h. The solvent was removed
in a stream of nitrogen. As an intermediate step the skyllamycin
sample was additionally hydrolyzed (1108C, 6 N HCl, 24 h). Finally
samples were derivatized and analyzed by GC-MS analytics accord-
ing to above mentioned conditions.
Experimental Section
Synthesis of b-hydroxy amino acids and of b-methyl aspartic
acid
Marfey derivatization and HPLC-MS analysis
100 mL of an aqueous 1% FDAA (1-fluoro-2,4-dinitrophenyl-5 l-ala-
nine amide) solution were added to a 50 mL aliquot of a 0.05m so-
lution of the amino acids or the skyllamycin hydrolysate in H2O.
Then 20 mL of 1 N NaHCO3-solution were added as a base and the
mixture was heated at 408C for 80 min. The solution was adjusted
to pH 7 adding 10 mL of 2 N HCl. After freeze-drying the residue
was dissolved in 1 mL of MeOH. The HPLC-MS analyses were per-
formed on a HPLC 1100 series (Agilent Technologies) hypenated to
a QTRAP 2000 ESI-Quadrupol-MS (Applied Biosystems). The sam-
ples were separated on a Phenomenex Luna C18 column (1ꢃ
50 mm) and eluted with 5% MeCN-0.1% HCOOH in H2O at a flow
rate of 60 mLminÀ1. The elution program was set as follows: 0–
58 min (5–40% MeCN), 58–59 min (40–100% MeCN), 59–62 min
(100% MeCN). Retention times of the FDAA derivatives (min): S-Thr
(25.1), S-aThr (26.0), S-Asp (26.4), R-aThr (28.5), R-Asp (29.0), S-Ala
(30.6), R-Thr (31.5), S-Pro (32.4), R-Pro (34.5), (2S,3S)-b-OH-Leu
(35.2), R-Ala (36.3), (2S,3R)-b-OH-Leu (37.1), (2R,3R)-b-OH-Leu (41.4),
(2R,3S)-b-OH-Leu (44.7), S-Trp (46.9), S-Leu (47.5), R-Trp (50.8), R-Leu
(53.8).
The four stereoisomers of b-hydroxy leucine,[13] b-hydroxy phenyla-
lanine[14] and two stereoisomers of b-methyl aspartic acid[15] were
synthesized according to literature procedures. Amino acids
(2S,3S)- and (2R,3R)-b-hydroxy-O-methyl tyrosine were synthesized
according to a strategy applying the Sharpless dihydroxylation re-
action.[16]
Hydrolysis of skyllamycin A and B
Skyllamycin (0.4 mg) was dissolved in 1 mL of 6 N HCl. The solution
was degassed and then heated under nitrogen atmosphere at
110 8C for 24 h. For Trp analytics the same amount of skyllamycin
was dissolved in 1 mL of 6 N HCl containing 3% w/v of phenol and
1% w/v of dithioerythritol, degassed and heated under nitrogen at
1208C for 24 h. Subsequently the hydrolysate was dried at 1108C
under a stream of nitrogen.
Derivatization of a-amino acids for chiral GC-MS analytics
To amino acid or the dry skyllamycin hydrolysate, 200 mL of 2 N
HCl in ethanol were added. The mixture was heated at 1108C for
30 min. Then reagents were removed at 1108C in a stream of nitro-
gen. Subsequently 100 mL dichloromethane and 50 mL trifluoroace-
tic anhydride were added and the mixture was again heated at
110 8C for 10 min. Reagents were removed at room temperature in
a stream of nitrogen thus rendering the N-trifluoroacetyl ethyl
esters for subsequent GC-MS analytics.
NOESY-NMR
NMR spectra were recorded on a Avance III 500 MHz NMR spec-
trometer (Bruker) equipped with a BBF probe head. NOESY spectra
were collected in [D6]DMSO, the mixing time was varied between
50 and 600 ms. The NMR data were processed using the program
TopSpin (Bruker), the cross-peak integrals were integrated using
the program Sparky (University of California, San Francisco, CA,
USA). Cross-relaxation rates were determined from the initial slope
of a polynomial fit (cubic polynomial) of the cross-peak integrals as
a function of the mixing time (build-up curves). To calculate the
proton–proton distances the distance between the protons 5 and
6 of Trp (dTrp5–6 =2.49 ꢂ) and its cross-relaxation rates were taken
as a reference.
Chiral GC-MS analysis
Chiral GC-MS analyses were performed on a Thermo GC 8000 Voy-
ager spectrometer with the chiral stationary phase Lipodex-E
(Machery & Nagel, length: 25 m, diameter: 0.25 mm). The dry deri-
vatization mixture was dissolved in 50 mL anhydrous toluene. For
Chem. Eur. J. 2014, 20, 4948 – 4955
www.chemeurj.org
4953
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Theoretical methods
Keywords: conformational
analysis
·
NOESY-NMR
The conformational analysis of skyllamycin was performed within
the molecular mechanic formalism. The topology of the standard
amino acids was generated from the Amber library[27] using the
ff99SB[28] force field. Concerning the non-standard amino acids,
a first optimization was performed at the density function theory
(DFT) level with B3LYP/6–31+G(d,p), using as initial geometries
the corresponding standard amino acids (e. g. using the S-Asp ge-
ometry for (2S,3S)-b-methyl aspartic acid). The topology was then
generated with the Antechamber program, using the ff99SB[28]
force field. The corresponding partial atomic charges were as-
signed within the Antechamber program and with the RESP
method from HF/6–31G* single-point calculations.[29,30] An explicitly
parameterized dimethylsulfoxide (DMSO) solvation model[31] was
used in order to approach the experimental conditions. Periodic
boundary conditions were applied in all directions. The molecular
systems were first minimized by the steepest descent method as
usually achieved to avoid unfavorable atomic contacts. The con-
tacts were then equilibrated during a 100 ps simulation in which
the system was heated to 300 K, employing Langevin coupling.
The equilibration step was followed by a 10 ns NPT simulation pro-
viding basis structures for all other calculations. The pressure was
kept constant at 1 bar, using isotropic position scaling.
spectroscopy · peptide antibiotics · skyllamycin · structure
elucidation
[1] R. H. Alvarez, H. M. Kantarjian, J. E. Cortes, Mayo Clin. Proc. 2006, 81,
1241–1257.
[2] J. Andrae, R. Gallini, C. Betsholtz, Genes Dev. 2008, 22, 1276–1312.
[3] L. N. Johnson, Quart. Rev. Biophys. 2009, 42, 1–40.
[4] S. Toki, T. Agatsuma, K. Ochiai, Y. Saitoh, K. Ando, S. Nakanishi, N. A.
Lokker, N. A. Giese, Y. Matsuda, J. Antibiot. 2001, 54, 405–414.
[5] S. Pohle, C. Appelt, M. Roux, H.-P. Fiedler, R. D. Sꢀssmuth, J. Am. Chem.
Soc. 2011, 133, 6194–6205.
[6] G. M. Sheldrick, P. G. Jones, O. Kennard, D. H. Williams, G. A. Smith,
Nature 1978, 271, 223–225.
[7] R. Ciabatti, J. K. Kettenring, G. Winters, G. Tuan, L. Zerilli, B. Cavalleri, J.
Antibiot. 1989, 42, 254–267.
[8] A. A. Tymiak, T. J. McCormick, S. E. Unger, J. Org. Chem. 1989, 54, 1149–
1157.
[9] H. Umezawa, S. Kondo, H. Iinuma, S. Kunimoto, Y. Ikeda, H. Iwasawa, D.
Ikeda, T. Takeuchi, J. Antibiot. 1981, 34, 1622–1624.
[10] R. Bhushan, H. Brꢀckner, Amino Acids 2004, 27, 231–247.
[11] G. Bunkꢅczi, L. Vꢄrtesy, G. M. Sheldrick, Angew. Chem. 2005, 117, 1364–
1366; Angew. Chem. Int. Ed. 2005, 44, 1340–4342.
[12] K. Meindl, T. Schmiederer, K. Schneider, A. Reicke, D. Butz, S. Keller, H.
Gꢀhring, L. Vꢄrtesy, J. Wink, H. Hoffmann, M. Brçnstrup, G. M. Sheldrick,
R. D. Sꢀssmuth, Angew. Chem. 2010, 122, 1169–1173; Angew. Chem. Int.
Ed. 2010, 49, 1151–1154.
[13] K. Makino, N. Okamoto, O. Hara, Y. Hamada, Tetrahedron: Asymmetry
2001, 12, 1757–1762.
[14] D. Crich, A. Banerjee, J. Org. Chem. 2006, 71, 7106–7109.
[15] S. Schabbert, M. D. Pierschbacher, R.-H. Mattern, M. Goodman, Bioorg.
Med. Chem. 2002, 10, 3331–3337.
[16] K. C. Nicolaou, C. N. C. Boddy, H. Li, A. E. Koumbis, R. Hughes, S. Natar-
jan, N. F. Jain, J. M. Ramanjulu, S. Brꢆse, M. E. Solomon, Chem. Eur. J.
1999, 5, 2602–2621.
[17] J. Shoji, H. Hinoo, T. Hattori, K. Hirooka, Y. Kimura, T. Yoshido, J. Antibiot.
1989, 42, 1460–1464.
To ensure a complete potential energy surface analysis of both sky-
llamycin isomers (R- and S-configured a-OH-Gly9), a 30-cycle simu-
lated annealing procedure was performed within the Berendsen
temperature coupling algorithm. Each cycle consisted of 1) a 1 ps
heating stage up to 2000 K, 2) a 2 ps equilibration stage at 2000 K,
and 3) a cooling stage from 2000 K to 0 K for 13 ps. The tempera-
ture bath was cooled linearly over this time. Improper and dihedral
angle restraints were applied to prevent the interconversion of ste-
reocenters at 2000 K. The explicit solvent was restrained to avoid
vaporization artifacts due to high temperature. For both R- and S-
configured isomers of a-OH-Gly9 of skyllamycin, simulated anneal-
ing calculations were carried out either without or with constraints
(derived from NOESY-NMR data) providing 60 (30 and 30, respec-
tively) conformations for each isomer. These 60 conformations
were systematically minimized following the steepest descent
method, before distance analysis.
[18] A. Zampella, R. D’Orsi, V. Sepe, A. Casapullo, M. C. Monti, M. V. D’Auria,
Org. Lett. 2005, 7, 3585–3588.
[19] The unconstrained simulated annealing did not allow a full exploration
of the potential energy surface, in particular in the close surrounding of
the experimentally d1 and d2 NOESY-NMR-derived distances. However,
together with those obtained with the NMR-constrained simulated an-
nealing, these data ensure a more complete sampling of the conforma-
tional space.
[20] The fully unconstrained free MD simulations were not able to provide
a sufficient sampling of the conformational space, see Supporting Infor-
mation for a detailed explanation.
As a complementary tool to validate the conformational analysis,
free 100 ns MD simulations (NPT) were performed for both skylla-
mycin stereoisomers (R-a-OH-Gly9 and S-a-OH-Gly9). Two types of
simulation were achieved for both systems, namely with and with-
out NMR constraints (during the first 10 ns). All molecular mechan-
ic calculations were performed with the AMBER10 package.[27]
[21] S. Uhlmann, R. D. Sꢀssmuth, M. J. Cryle, ACS Chem. Biol. 2013, 8, 2586–
2596.
[22] J. W. Trauger, C. T. Walsh, Proc. Natl. Acad. Sci. USA 2000, 97, 3112–3117.
[23] K. Hoffmann, E. Schneider-Scherzer, H. Kleinkauf, R. Zocher, J. Biol.
Chem. 1994, 269, 12710–12714.
Acknowledgements
[24] S. T. Prigge, R. E. Mains, B. A. Eipper, L. M. Amzel, Cell. Mol. Life Sci. 2000,
57, 1236–1259.
[25] T. Kawahara, K. Suzuki, Y. Iwasaki, H. Shimoi, M. Akita, Y. Moro-oka, Y.
Nishikawa, J. Chem. Soc. Chem. Commun. 1992, 625–626.
[26] Y. Umeda, M. Moriguchi, K. Ikai, H. Kuroda, T. Nakamura, A. Fuji, T. Takeu-
chi, H. Umezawa, J. Antibiot. 1987, 40, 1316–1324.
[27] Case, D. A., Darden, T. A., Cheatham, III T. E., Simmerling, C. L., Wang, J.,
Duke, R. E., Luo, R., Crowley, M., Ross C. Walker, Zhang, W., Merz, K. M.,
Wang, B., Hayik, S., Roitberg, A., Seabra, G., Kolossvꢇry, I., Wong, K. F.,
Paesani, F., Vanicek, J., Wu, X., Brozell, S. R., Steinbrecher, T., Gohlke, H.,
Yang, L., Tan, C., Mongan, J., Hornak, V., Cui, G., Mathews, D. H., Seetin,
M. G., Sagui, C., Babin, V. and Kollman P. A. 2008.
[28] W. D. Cornell, P. Cieplak, C. I. Bayly, I. R. Gould, K. M. Merz, D. M. Fergu-
son, D. C. Spellmeyer, T. Fox, J. W. Caldwell, P. A. Kollman, J. Am. Chem.
Soc. 1995, 117, 5179–5197.
VS and RS were supported by the Cluster of Excellence
(UniCat) funded by the German Research Council (DFG). FDM
and PT thank the “Conseil Rꢄgional du Limousin” for financial
support and CALI (CAlcul en LImousin) for computing facilities.
The authors acknowledge the COST action 0804 “Chemical
Biology with Natural Compounds”). PT gratefully acknowledges
the support by the Operational Program Research and Devel-
opment for Innovations-European Regional Development Fund
(project CZ.1.05/2.1.00/03.0058 of the Ministry of Education,
Youth and Sports of the Czech Republic). PLS gratefully ac-
knowledges the Alexander von Humboldt-Foundation for
a postdoctoral fellowship.
Chem. Eur. J. 2014, 20, 4948 – 4955
www.chemeurj.org
4954
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[29] J. Wang, W. Wang, P. A. Kollman, D. A. Case, J. Mol. Graphics Modell.
2006, 25, 247–260.
[30] J. Wang, R. M. Wolf, J. W. Caldwell, P. A. Kollman, D. A. Case, J. Comput.
Chem. 2004, 25, 1157–1174.
[31] F. Y. Dupradeau, A. Pigache, T. Zaffran, C. Savineau, R. Lelong, N. Grivel,
D. Lelong, W. Rosanski, P. Cieplak, Phys. Chem. Chem. Phys. 2010, 12,
7821–7839.
Received: November 21, 2013
Published online on March 12, 2014
Chem. Eur. J. 2014, 20, 4948 – 4955
www.chemeurj.org
4955
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim